DYNAMIC RESPONSE AND ACOUSTIC FATIGUE*

OF STIFFENED COMPOSITE STRUCTURE

J. Soovere

Lockheed-California CompanyBurbank, California, USAA -

1. INTRODUCTION

Aircraft structure made from graphite/epoxy (Gr/E) composites offer asignificant weight savings over comparable aluminum alloy structures. Theinitial application of these composites has been restricted to low load-carryinglightweight structures such as ailerons, flaps, elevators, slats, and fairingsto gain in service experience at low risk. The Gr/E composites may also be usedin lightly loaded fuselage structures on some vertical/short takeoff and landing(V/STOL) aircraft. For minimum weight, the fuselage structure will be requiredto operate well into the post-buckling region, with initial buckling justabove 1g.

These lightweight structures are very often subjected to a high level jetnoise environment, especially during takeoff and landing, which can produce highrandom vibration stresses in these structures. High cycle fatigue failures,commonly referred to as acoustic fatigue failures, have occurred in thesestructures [i] , [2] as a result of the random vibration stresses. The V/STOLfuselage can also be subjected, simultaneously, to a thermal environment. Otherenvironmental factors such as moisture penetration, impact damage and lightningstrikes may degrade the performance of these structures in a high noiseenvironment.

The compression and shear buckling loads experienced by the fuselagestructure introduce peel loads at the skin-stiffener interface similar to thoseproduced by jet noise. Initial studies with Gr/E composites [3] indicated thatfasteners provided an improved acoustic fatigue life over structural bonding.For this and other practical reasons, initial application of Gr/E composites totransport aircraft, such as the L-1O11 composite aileron [4] , [5] in Figure 1,used mechanical fasteners. Bonding has also been used on some experimenLal Gr/Eacoustic fatigue panels [3] , [6]. Integrally stiffened Gr/E :anels [7] , [8] ,[9] , [10] , where the skin and stiffeners are cocured, represent the next stagein this development. The peel strength and, Lherefore, the acoustic fatigue lifeof these integrally stiffened structures can be further improved by the use ofstitching [8] , [11].

This paper summarizes the results obtained from acoustic fatigue and dynamicresponse tests and the L-1011 composite aileron and integrally stiffened a+JEpanels. The nature of the damping in integrally stiffened composite panels, itstheoretical prediction and its implication on internal noise are brieflydiscussed.

2. COMPOSITE AILERON TEST PROGRAM

The composite aileron was fabricated with a minisandwich skin [4] , [5] inwhich the Gr/E face sheets were separated by a syntactic (SYNT) core containing _. .glass microballoons. Cocured doublers were located on the inner face sheet atthe rib locations. The covers were attached to the Gr/E cloth ribs, and thefront and rear spars with fasteners.

*Performed in part under Contracts N62269-80-C-0239, NADC, and NAS 1-15069, NASA,and in part with Lockheed-California Company funding. •

The composite aileron test program included the development of randomfatigue data by means of double and single cantilever coupons representing theskin-spar and the skin-rib interfaces and the rib bend radius. Typical measuredstrain distributions [4] are illustrated in Figure 2. The results of thesecoupon random fatigue data [5] , including the effects from a built-in voidadjacent to a fastener head, a single impact damage (0.88 kilogram-meter) 0 0adjacent to a fastener head, moisture conditioning and elevated temperature(82'C), are summarized in Figure 3.

In the early design, core compression was experienced along the edges ofthe doubler [4] , [5] which produced a premature separation of the doubler fromthe skin along these edges at lower random strain levels (Figure 3). Improvements 0 0in the fabrication procedure eliminated this problem [5]. Consequently, therandom fatigue data (Figure 3), which involved failures in the outer face sheetsadjacent to the countersunk fastener heads, were considered to be more represen-tative of the improved design.

Modal studies conducted on a representative section of the composite aileron, 0 0using impedance head hammer tap and loudspeaker excitations, indicated very lowdamping ratios (lable 1) even after damage [5] from simulated lightning strikes(Figure 4). The damage was mostly confined to the visible surface damage areawith very little interlaminar damage. All of the modes dropped in frequency bya small amount, reflecting, in view of the amount of damage sustained, theredundant nature of the composite structure.

The random fatigue data were used, in conjunction with the results of anonlinearity test, to select the accelerated proof test random spectrum level(Figure 5). Nonlinear panel response (Figure 6) was obtained during the prooftest. The measured strains, when compared with the random fatigue data, indicatedthat the composite aileron would be free of acoustic fatigue failures throughout 0 0its design life.

3. COMBINED ACOUSTIC AND SHEAR LOADS

Acoustic fatigue tests were conducted on an integrally J-stiffened Gr/Eminisandwich (Figure 7) and a monolithic (Figure 8) panel near initial shearbuckling [7] , [10]. Both integrally stiffened panels were designed to aninitial shear buckling load of 1786 kilograms per meter (Table 2). The analysiswas performed with an anisotropic finite element program (STRAP 5) developed bythe Lockheed-California Company. In general, good agreement was obtained betweenthe predicted and measured buckling loads for the three bay monolithic panel andsimilar four bay monolithic panels tested in another program [12]. The toolingused affected the thickness of the minisandwich panel skin which resulted in ahigher measured initial buckling load. Good agreement was obtained between thepredicted and measured mode shapes, the latter obtained from Moire fringe pat-terns measured in the test facility illustrated in Figure 8. The monolithic panelbuckled mode shape contained five antinodes in each of the three bays whereas theminisandwich panel exhibited two antinodes only in the large center bay. S

The initial buckling was measured with a noncontacting displacementtransducer located aL an antinode. The initial buckling load was determinedfrom the discontinuity in the load displacement curve [7] , [i] , and from

the frequency-load curve (Figure 9). The increase in the damping ratio(Figure 10) in both the critical and noncritical modes, on approaching initial - Sbuckling, made it difficult to trace the variation of the modal frequenciesthrough buckling with the impedance head hammer tap method. The higher thecritical mode number, the greater the difficulty. This increase in damping onapproaching buckling, was previously [13] observed in axial compression testson stiffened aluminum panels. The damping was also found to be nonlinear in thecompression load region. The same result can also be expected for stiffenedcomposite panels under axial compression load.

In spite of the high damping near initial buckling, the overall strain wasincreased by approximately thirty percent (Figure 11) at the critical locationon both panels, when excited with broad band random acoustic loading. Nonlinearresponse (Figure 12) was obtained in both panels at the higher sound pressurelevels.

4. COMBINED ACOUSTIC AND THERMAL ENVIRONMENTS

The effect of a 121 0C thermal environment on the sonic fatigue life ofintegrally stiffened Gr/E panels, representative of potential fuselage structure,was investigated [7] , [9] using two advanced J-stiffened monolithic panels andtwo advanced blade stiffened orthogrid panels, illustrated in Figures 13 and 14, 0

respectively. These panels were also designed to an initial shear buckling loadof 1786 kilorams per meter (Table 2). The panels were mounted, in turn, intoa steel test frame in which the thermal expansion was matched biaxially to thatof the composite panel. In the modal studies, heat was applied to the outersurface of the panels by six infrared lamps and the excitation was provided byimpedance head hammer taps at preselected grid points. The excitation and the 0corresponding displacement were analyzed within the Hewlett-Packard HP 5451CFourier Analyzer to obtain the resonant frequencies, damping ratios and modeshapes.

The measured fundamental mode shape for both panels is illustrated inFigure 15. The fall-off in the resonant frequencies with temperature (Figure 16),observed for both panel configurations, is due to thermally induced bi-axialcompression loads in the skin, introduced by differences in the thermal expansionof the Gr/E frames and skin because of differences in their fiber orientations.The temperature does not appear to have affected the damping in both the mono-lithic and the orthogrid panels (Table 3). The fundamental mode damping ratioin the orthogrid panel is, however, an order of magnitude greatcr than that inthe monolithic panel. The damping in the higher order panel modes is morecomparable and very low for both panels.

One panel of each design was acoustic fatigue tested at ambient temperatureand the other at a temperature of 123 0C. Heat was supplied by a speciallydesigned quartz lamp heater panel, mounted inside the progressive wave tunnel. 0An overall sound pressure level of 167 dB (Table 4) was used to fail the roomtemperature orthogrid panel (Figure 14). In contrast, the room temperature andelevated temperature monolithic panels were failed with an overall sound pressurelevel between 160.8 to 164 dB. The difference in the sound pressure levelrequired to produce the same long side rms strain (Table 4) in the orthogridand monolithic ambient temperature panels can be attributed to the effect S

expected from the differences in their fundamental mode damping, even in thepresence of nonlinear panel response (Figure 17). The failure obtained withthe orthogrid panel does not reflect the true capability of the design sincethese panels contained stress concentrations at the ends of the longitudinalstiffeners (Figure 14) where the failure was initiated.

In spite of some indications that the temperature of 123 0C could affect thesonic fatigue life, the results are considered inconclusive on account of thesmall sample size. The current room temperature semi-empirical analysis method[6] did not predict the rms strains in the room temperature panels with any

0 degree of accuracy (Table 4). The main reason is thought to be the omissionof damping in the above method although differences in the spectrum shape ofthe random noise may also be a contributing factor.

0 0 i*- -

5. RANDOM FATIGUE DATA FOR INTEGRALLY STIFFENED Gr/E PANELS

Random fatigue data were developed with double cantilever coupons (Figure 18),representing the skin-longitudinal stiffener interface, for the integrallystiffened monolithic (Figure 19) and orthogrid (Figure 20) panels at room temp-erature. The monolithic coupon data exhibited higher rms strain (Figure 19)levels than the bonded panel data in reference [6]. The monolithic panels couldwithstand even higher strains than the corresponding coupons. Stitching(Figure 18), also illustrated in cross-section in (Figure 21), increased the

failure strain level by approximately 70 percent, comparable to the improvementmeasured in static peel tests [Ii].

The orthogrid coupons (Figure 20) achieved an rms strain level comparableto that measured in the composite aileron skin at a KT of 1, and that achievedwith stitching in the monolithic coupon. The composite aileron and the orthogridpanel had the same fiber orientation in the face sheets. The orthogrid panelfailed well below the coupon strain level indicating the magnitude of the stressconcentration at the ends of the axial stiffeners.

6. THEORETICALLY PREDICTED DAMPING FOR THE ORTHOGRID PANELS

Fastener related damping has been eliminated in integrally stiffened com-posite panels while the material damping is very small, usually below a viscousdamping ratio of 0.001 for Gr/E composites. Consequently the only source ofsignificant damping is acoustic radiation, once the energy loss to the surroundingstructure has been minimized. A characteristic of acoustic radiation is that thedamping is high in the fundamental mode and falls off with increase in modenumber due to the cancellation effect [14]. Furthermore, the acoustic radiationis proportional to panel area and should be very large in the fundamental modeof a large single panel. These conditions have been met by the damping ratiosin Table 3. The predominant contribution of acoustic radiation to the dampinghas also been confirmed with stiffened composite honeycomb panels [15]. Theorthogrid panel represents an ideal example for demonstrating the above conclu-

sion. The simplified expression in reference [14], for the viscous damping ratio

produced by acoustic radiation from a simply supported panel, was used. Themeasured and predicted damping ratio in Figure 22 are in reasonable agreement,particularly for the fundamental mode, when considering the simplicity of theanalysis.

7. CONCLUSIONS

The importance of developing nonlinear analysis capability for designpurposes has been demonstrated since composite structujres can sustain highvibration levels, in the nonlinear response region, over a wide range of com-bined loads and environmental conditions without failure. All of the randomfatigue data exhibited a fatigue limit just beyond 107 cycles, an importantconsideration for the design of acoustic fatigue resistent composite structures.The acoustic fatigue resistance in integrally stiffened composite structures canbe considerably improved by attention to detailed design at the critical loca-

tions. In this respect, the integrally stiffened panels are considered to besuperior to secondary bonded panels. Since the damping in integrally stiffenedcomposite panels is due to acoustic radiation, the transmission of turbulentboundary layer noise, through the lighter weight integrally stiffened compositefuselage structure, could be significantly increased, over current fuselagestructure, for the same density of acoustic sidewall treatment. A compensatingincrease in the sidewall treatment density could eliminate much of the weight-saving achieved with composite fuselage structure.

0 0

REFERENCES

1. T.R. RIEBEN 1957 WADC-TR-57-513. Sonic Fatigue Testing and Development

of Aircraft Panels.

2. E.J. RICHARDS 1960 WADC-TR-59-676. An Introduction to Acoustic Fatigue. 0 •

3. M.J. JACOBSON 1972 AFFDL-TR-71-126. Advanced Composite Joints; Designand Acoustic Fatigue Characteristics.

4. J. SOOVERE 1980 The Shock and Vibration Bulletin, No. 50, Part 4. SonicFatigue Testing of NASA L-1011 Composite Aileron. 0

5. J. SOOVERE 1982 Journal of Aircraft, 19, No. 4. Sonic Fatigue Testingof an Advanced Composite Aileron.

6. I. HOLEHOUSE 1980 AFWAL-TR-80-3019. Sonic Fatigue Design Techniques for ...Advanced Composite Aircraft Structures. 0

7. J. SOOVERE 1982 NADC-78169-60. Effect of Acoustic, Thermal and ShearLoading on Flat Integrally Stiffened Graphite/Epoxy Fuselage Panels.

8. M.J. JACOBSON 1981 NADC-81045-60. Fatigue of V/STOL Composite Fuselage .Panels Under Acoustic-Thermal Environments. 0 •

9. J. SOOVERE 1983 24th Structures, Structural Dynamics and MaterialsConference, Lake Tahoe, Nevada, USA. The Effect of Acoustic-ThermalEnvironments on Advanced Composite Fuselage Panels.

10. J. SOOVERE 1983 Second United States/Japan Conference on CompositeMaterials NASA-Langley Research Center, Hampton, Virginia, USA. DynamicResponse of Flat Integrally Stiffened Graphite/Epoxy Panels Under CombinedAcoustic and Shear Loads.

11. J. SOOVERE 1980 LR 29410; 1982 LR 30126; Lockheed Reports. AcousticResponse of Composite Panels Under Combined Loading - Parts I and 2.

12. R.B. OSTROM 1981 NADC-78137-60. Post-Buckling Fatigue Behavior of Flat,Stiffened Graphite/Epoxy Panels Under Shear Loading.

13. J. SOOVERE and S.T. CHIU 1976 AFFDL-TR-76-60. Effects of CombinedAcoustic and Flight Loads on Crack Growtb 0

14. D.J. MEAD 1964 AFML-TR-65-z84. The Effect of Certain Damping Treatmentson the Response of Idealized Aeroplane Structures Excited by Noise.

15. J. SOOVERE Ph.D. Thesis, University of Southamptun. Dynamic Response of -

Acoustically Excited Stiffened Composite Honeycomb Panels (In Preparation).

* 0

TABLE 1. MODAL FREQUENCIES AND DAMPINGS

AFTER SIMULATEDUNDAMAGED AILERON LIGHTNING STRIKES

VISCOUS DAMPING VISCOUS DAMPINGFREQUENCY, Hz RATIO FREQUENCY, Hz RATIO

96.4 0.0042 91.18 0.0063

125.9 ... 109.2 0.0038

134.3 0.0040 119.2 0.0067

149.0 0.010 126.6 0.0065

129.7 0.0091

TABLE 2. INITIAL SHEAR BUCKLING LOAD

NXYCR Kg/M

TYPE OF LENGTH WIDTH THICKNESS STRAP TESTPANEL mm mm mm SKIN LAYUP 5 AVEPAGE

J-Stiffened 546 152 0.89 (±45/0/90)s 1821* 1759**

Monolithic 533 180 1.02 (45/0/-45/90) 1893* 1789 -

s 1841

J-Stiffened 2419 -Misnd 533 304 1.52 (45/O/-45/SYNT) 1839* 2682Minisandwich s2682

AdvancedJ-Stiffened 737 180 1.02 (45/0/-45/90)s 1839* --Monolithic

Advanced 660 254 1.40 (45/0/-45/SYNT) 1839* --

Orthogrid s

* Target Value 1786 Kg/M

** Reference [12] Subscript s means symmetrically laminated.

.............. ..... ..... * .... . .

0 0

0-4

cI 0 o o

0~~ 000 I__

44 0 Ci244 4

o. o4 0 0 0Z 4

g! 2- - 00 0

E 4. Ol =' N 41

4(- .0 -0 I Z 0 C, C,

>z InC000z.

E-4.- N ( _ _ _ __-

-4 m CIS 004 ... E- 4 N CO CO In I0 ZO ('c 44- 0 'C I C' C) P'

000O'C L 'C C 'o

0, 0o 4m 0.14-404 m' m4 0 1 0 0Q -

HE, ;2 S41 'T 4

('4 N C' C H - 0 X 44 4

H C4U3 0

C17:I- 4 44 04 C 0 ca 4

0 000 04 0 _

CO4.4C CO 0 '4 '

E. 044 . . 0 0 0 (4 0 '

o ' 000 00

-4 z- 44 040 C' In4 a, C' '4 44

o~ 00 o'W 44 - 'Cj(

0 0 4

M 00 W0

HCO ZcO COO, v1 ' 'Co

4-444 W14 NOr - C 0 ' C C

E SUEI E ELEVEL 2

400O ELEVEL 2 E

"300

;4 C 00 -LEVL11T 00[2EVEL 1

0100 \Z2

5 4321 4STRAIN GAGE LOCATIONS STRAIN GAOL LOCATIONS

Fni

[a) SKIN.RIB COUPON (b) RIB BEND RADIUS COUPON

Figure 2. Typical Measured StrainFigure 1. L-1011 Composite Inboard DistributionAileron with the Lower Cover Removed

AWDs AS OAS- AS 921 $AS 641 OAS

SKIN COUPONS OUTIOD lo2 111 67

0 CORE FAILURE KT- 1SKINRIB COUPONS $RONT SPAR

0 UPPER SKIN FAILURE AT FASTENER HEADO DOUBLER SEPARATION+ OUBLER SEPARATION - 12 7mm OIA, VOID

* DOURILER SEPARATION - 82-C AND MOISTUREA DAMAGE GROWTH - 0 88Kg M IMPACTA OAMAGE GROWTH - 0 88Kg M IMPACT, 82 -C

AND MOISTURE 3

200 I I a

RAR $PAR

SRAILING EDOEE 1500 t "

E

1000 SWEPT STROKE AT MAIN RIB

10 0,IREC T STROKE AT PANEL CENTER0 4 RSWEPT STROKE AT PRONT SPAT

%" DIRECT STRIKE AT REAR SPAR. ......... 6 SWEPT SIAOKE AT INTERMEDIATE RIOS500-

c Figure 4. Exterior of Composite AileronI After Lightning Strike Tests

105 106 107 108CYCLES

Figure 3. Random Fatigue Data for the 0

Skin and the Skin/Stiffener InterfaceANALYSIS BANDWIDTH 0.244 Hz

NUMBER OF AVERAGES 800

160.10FPROOF PROOF

* TEST SPECTRUM /TEST

• 150. _ L-1011 DESIGN

W - I I-

UJLU z

1 30 .KG

W c*)" 120. 0.

AIRz cc -- I

u NOISE

100.d

KG110.ARNOS10B

100. 50 100 250 500 1000

FREQUENCY - Hz 110 120 130 140 150 160 170FREQUENCY - Hz

Figure 5. Inboard Aileron Design,Proof Test, and Air NoiseProTe, Band Aircta NFigure 6. Narrow-Band Analysis of1/3-Octave Band Spectra Aileron Strain Data

Figure 7. Reverse Side of Shear Frame

Showing Stiffener Pivot Attachments -Figure 8. Load Frame in Acoustic"

J-Stiffened Minisandwich Panel Progressive Wave Tunnel Wall Configured

for Shear Load - J-Stiffened

Monolithic Panel

._ 08

R EGD EN 3 1 0 7 " - 1 1

200 (CR,,,ICAL) P 3 o1z z 05

a.... .

uW MVODE I I

100. . . 04

_0 0201000 2000 1000,> .,.. .

LOAD - Kg 01,- BUCKLING+ LOAD

1 1000 2000 3000Figure 9. J-Stiffened Minisandwich LOAD- Kg

Panel Frequency Variation as aFunction of Jack Load Figure 0. Variation of Damping withf S Jack Load for J-Stiffened

Minisandwich Panel

100100 Hz ANALYSIS BANDWIDTH 60 AVERAGES

-- 16 di OVERALL

-- JACK LOAD Kg N_ J tOISE LEVEL

0 0 - 1--0

*0- 100 140 d8 OVERAL

M EOIS

o (3

00

0 JACK

OLOAD KA

145 150 155 160 165 1,

OVERALL SOUND PRESSURE LEVEL - dB 0 010 50 100 150 200 250 300 350 400 450 500

fREiUENCY - H

F toFigure 1i. Variation of OverallFoSound Pressure Level - J-Stiffened Figure 12. Strain Power Spectral Density

Monolithic Panel for J-Stiffened Minisandwich Panel

0 4

* 4

Figure 14. Acoustic Failure in BladeStiffened Orthogrid Graphite/Epoxy Panel

Figure 13. J-Stiffened MonolithicGraphite/Epoxy Panel

(a) Monolithic Panel (b) Orthogrid Panel

Figure 15. Measured PanelFundamental Mode Shape

6 500 T T - - -- 1 Hz ANALYSIS BANDWIDTH 60 AVERAGES SS100000

~157.3 do OVERALL

400 FODE 10000

- ------- 4,1 1000

3,1 0-, 100

200!

3: 142 8 dB OVERALL

too RP 1, NOtSELEVEL0 __~~______________0 0.10 50 100 150 200 250 300 3S0 400 400 00

50 100 150 FREQUENCY - Hz

TEMPERATURE -°C

Figure 17. Typical Strain Power

Figure 16. Variation of Modal Spectral Densities forFrequencies with Temperature- Monolithic Panel

Monolithic Panel

0 O

- - Figure 18. Unstitched and StitchedJ Coupons Mounted on Electromagnetic S

Shaker

*A INTEGRALLY STIFFENED GRJE PANEL 0- NO FAILURE OBTAINED

-- 0-- -BONDED GR/E PANELS |6] 0 STITCHED INTEGRALLY *INTEGRALLY STIFFENED GRIE COUPONS STIFFENED GR'E COUPON

1500FAILED AFTER2 w05 CYCLES

1000- AT THIS LEVEL NO FAILURE

00

0-. 0 >

3x 105 106 107 3x 107NUMBER Of CYCLES

Figure 19. Ambient Temperature RandomFatigue Data from Monolithic 0

Panels and Coupons

2 0 00 -• , I

E 1500 0 0A

B oK 1000 -

--- ORTHOGRID COUPONS

5 Soo A ORTHOGRID PANEL0 L 1011 COMPOSITE AILERON FACE

SHEET COUPONS IKT 1) 141.15)

0 I i , ]1 I I n I II 1 t

3 x 105 16 in7 3 x 107NUMBER OF CYCLES

Figure 20. Ambient Temperature RandomFatigue Data from Orthogrid

Coupons and Panel

S' 0

-~, -'sb- z, z,

Alt

Figure 21. Sectional M11icrographThrough Stiffener Flange and Panel

Along the Stitch Line-0X25 Magnification

PANELNO TEMPERATURE

0 01 AMBIENT904- @01 123 0 C10 0

002 AMBIENT 11cc NUMBERS REFER

.0 TO MODES03

w 02 I

a02 2LU

43 0

4 4

0 01 02 03 04 05MEASURED DAMPING RATIO

Figure 22. Comparison of Predictedand Meqsixred Orthogrid Panel Damping 5